Machine learning-based risk prediction model for arteriovenous fistula stenosis

• DOI: 10.1186/s40001-025-02490-x
• Publication Date: 2025-03-29T18:23:56Z
• Authors: Peng Shu; Ling Huang; Shanshan Huo; Jun Qiu; Haitao Bai; Xia Wang; Fang Xu

Arteriovenous fistula stenosis is a common complication in hemodialysis patients, yet effective predictive tools are lacking. This study aims to develop an interpretable machine learning model for risk prediction. Clinical data from 974 patients (55 features) undergoing arteriovenous dialysis at The Central Hospital of Wuhan (2017-2024) were analyzed retrospectively. dataset was split into training (70%) and test (30%) sets. Seven models-Random Forest, XGBoost, Support Vector Machine, Logistic Regression, K-Nearest Neighbors, Artificial Neural Network, Decision Tree-were trained. Performance evaluated using F1 score, accuracy, specificity, precision, recall, AUC-ROC. SHAP values identified key predictors the optimal model. XGBoost achieved highest AUC (0.829, 95% CI 0.785-0.880). analysis highlighted seven critical predictors: number surgeries, prothrombin time activity, lymphocyte count, duration, triglycerides, vitamin B12, C-reactive protein. effectively predicts clinical data. explanations enhance interpretability, aiding personalized care strategies.